DROZlegal / Blog / Can AI Give Legal Advice in Ontario?

Can AI Give Legal Advice in Ontario?

No. Under Ontario's Law Society Act, only a licensed lawyer or paralegal may practise law or provide legal services — Section 26.1(1) says plainly that "no person, other than a licensee whose licence is not suspended, shall practise law in Ontario or provide legal services in Ontario," and an AI tool isn't a licensee. That doesn't make AI risk-free, though: the Law Society of Ontario's own guidance warns that a chatbot interacting directly with the public risks "inappropriately or inadvertently providing legal advice," and when that happens, the exposure lands on the lawyer or firm running it, not on the software.

This is general information, not legal advice, and reflects Ontario law as of September 2, 2026.

Why the AI itself isn't “practising law”

Ontario restricts who may practise law to a narrow group. Section 26.1(1) of the Law Society Act says a person other than a licensee in good standing may not practise law or provide legal services in Ontario. The Act's own definition of "legal services" (s. 1(5)-(6)) turns on a person applying legal principles and legal judgment to someone's circumstances or objectives — giving advice about a person's legal rights or responsibilities is one of the examples it names directly. A large language model predicting the next plausible word in a sentence isn't a person, and it can't hold a licence. Read literally, the prohibition governs who may practise, not which software they use while doing it.

Where the exposure actually lands

That's not the Law Society of Ontario's blessing for client-facing AI, though. Its April 2024 white paper on generative AI names the real risk directly: when a chatbot deals directly with the public, "there is a risk that it may be inappropriately or inadvertently providing legal advice… or creating other misunderstandings." The same guidance is just as direct that certain tasks "will require the input of the licensee and the application of their legal competence and cannot be delegated." Put the two together and the practical rule is simple: the software never becomes the one practising law without a licence — the lawyer or firm that deployed it unsupervised does, by letting a tool stand in for judgment a licensee was supposed to apply.

What stays with a human either way:

  • Verification. Reading and confirming anything AI drafts before it reaches a client or a filing.
  • Judgment calls. Applying legal principles to a client's specific circumstances — the part of "legal services" the Act actually defines.
  • Supervision. Checking any AI-assisted work from staff the same way you'd check their non-AI work.

That's the same competence and supervision framework our deeper look at the LSO's generative-AI guidance walks through in full, mapped to the specific professional-conduct rules behind it. If you're evaluating a specific AI vendor for your practice, our four-question vendor safety checklist is the practical next step before anything client-facing goes live.

Practising law without a licence is one question; a separate one is whether a lawyer has to tell the court AI touched a filing at all. See our explainer on whether Ontario lawyers must disclose AI use in court filings for what the actual rule requires versus what it doesn't.

Newsletter

Read this before your next reconciliation.

Compliance notes and product updates for firms that would rather not find out about a rule change the hard way.

Get started

See where a lawyer's sign-off never gets automated away

Built around the same human-judgment principle this answer describes — work advances, but a lawyer approves before anything reaches a client.

Free: the complete guide to AI for Canadian law firms → Lawyer AI Academy